Applications

Distilling Free-Form Natural Laws from Experimental Data

AI Feynman 2.0: Pareto-optimal symbolic regression exploiting graph modularity

Learning to Infer Graphics Programs from Hand-Drawn Images

Making sense of sensory input

Programmatically Interpretable Reinforcement Learning

Symbolic methods

Bootstrap Learning via Modular Concept Discovery

Meta-interpretive learning of higher-order dyadic datalog: predicate invention revisited. Code. Web interface.

Neurally-guided program induction

DeepCoder: Learning to Write Programs

RobustFill: Neural Program Learning under Noisy I/O

Execution-guided Neural Program Synthesis

Write, Execute, Assess: Program Synthesis with a REPL

Accelerating search-based program synthesis using learned probabilistic models

Learning Differentiable Programs with Admissible Neural Heuristics (This paper is hard to categorize. Neural networks are used as a relaxation of the space of programs, and this relaxation serves to construct a search heuristic. But definitely worth reading.)

Neural-symbolic programs

Making Sense of Raw Input

Learning to learn generative programs with Memoised Wake-Sleep

Imitation-Projected Programmatic Reinforcement Learning

HOUDINI: Lifelong Learning as Program Synthesis

Modular meta-learning

Programs and gradient descent

Neural Turing Machines

Learning Explanatory Rules from Noisy Data

TerpreT: A Probabilistic Programming Language for Program Induction